Let’s say you have a graph that you want to visualize. This graph has several attributes, including locations of the nodes, and weights of the edges. Many common graph visualization tools do not allow you to provide coordinates for the nodes, and many others only plot in 2D. Thankfully, igraph in R exploits the rgl package and uses GL graphics to overcome this. Let’s begin:

As mentioned, we need igraph for this, so let’s load it.

library(igraph)

Now let’s load our graph and extract the attributes of interest, putting them into a format that we can use to plot them. This graph happens to be a connectome (map of the brain).

g <- read.graph('./demo_graph.graphml', format='graphml')
centroids <- get.vertex.attribute(g, 'centroid')
print(centroids[1:5])
## [1] "[110, 101, 74]" "[138, 137, 81]" "[98, 72, 98]"   "[124, 81, 115]"
## [5] "[99, 94, 93]"

We notice that in this case, the centroids we would like to plot are stored as strings. The format igraph expects is an Nx3 matrix, where N is the number of nodes in your graph. Therefore, we must do some string acrobatics to get the centroids into the format we want.

layouts <- matrix(nrow = length(centroids), ncol = 3)
for (i in 1:length(centroids)) {
  temp <- na.omit(as.numeric(unlist(strsplit(unlist(centroids[i]), "[^0-9]+"))))
  layouts[i,] <- temp
}
print(layouts[1:5,])
##      [,1] [,2] [,3]
## [1,]  110  101   74
## [2,]  138  137   81
## [3,]   98   72   98
## [4,]  124   81  115
## [5,]   99   94   93

Great! Now we can assign this value to our layour, and plot our graph. We set several attributes here, including changing vertex color, size, and turning off labels. We also change the color of our edges, and set their width to be proportional to the edge weights in our graph.

g$layout <- layouts

rglplot(g, layout = g$layout,
        vertex.label = NA, vertex.size = 12, vertex.color='tomato1',
        edge.width = log10(get.edge.attribute(g, 'weight')), edge.color='paleturquoise1')

unnamed_chunk_4snapshot
You must enable Javascript to view this page properly.

If we also wish to save our graph we can use the following block to save it as a pdf: rgl.postscript('./mygraph.pdf', fmt = 'pdf')